Object size estimation with industrial robot gripper using neural network and machine learning

This paper describes a research in finding and improving a method for object size estimation by the means of electric gripper and without visual aid in industrial robotic settings. Several models for size prediction utilizing machine learning of a neural network are investigated. The best performing model is chosen for testing its practical usability in pick-and-place sorting tasks. The used methodology as well as the resulting achievements are registered and presented. Finally, the degree of success is analyzed and future steps of improvement are proposed.